Pixel-based Image Analysis Research Articles

Assessing Two Decades of Land Use/Land Cover Changes in the Uluabat Lake Ramsar Site using Multi-Temporal Satellite Imagery

The Ramsar Convention on Wetlands designates over 2,000 sites of international importance, providing crucial habitats for diverse species. Uluabat Lake, faces anthropogenic pressures such as urbanization, agriculture, and industrialization, affecting its ecological integrity. Using multi-temporal Landsat 7 and Landsat 9 satellite images from 2002 and 2022, along with 2019 management plans, we assessed land use/land cover (LULC) changes in the lake's catchment area. Data were pre-processed with ENVI and stored in ERDAS Imagine. We employed pixel-based image analysis with maximum likelihood classification (MLC) to generate LULC maps and evaluated classification accuracy using ground truth data and the kappa coefficient. Our findings revealed a 15.8% reduction in lake area, from 136.1 km² in 2002 to 114.5 km² in 2022, primarily due to sediment transport from surrounding agricultural land and tributary streams. Urban-agricultural and reed-swamp areas increased by 74.7% and 59.6%, respectively, while shrubs and forests declined by 35.64%, largely from reed conversion to agriculture in the Mustafakemalpaşa River delta. Overall classification accuracy ranged from 88.2% to 91% with a kappa coefficient of 0.81 to 0.82. These transformations highlight the increase in reed and swamp areas and the decrease in lake area, emphasizing the need for effective conservation and management practices.

Comparing Object-Based and Pixel-Based Machine Learning Models for Tree-Cutting Detection with PlanetScope Satellite Images: Exploring Model Generalization

Despite utilizing various remote sensing datasets, precise tree-cutting detection remains challenging due to spatial and spectral resolution constraints in satellite imagery, complex landscapes, data integration issues, and the need for accurate multi-temporal reference datasets. This study investigates the utilization of PlanetScope (PS) satellite images, along with pixel-based (PBIA) and object-based (OBIA) image analysis, for accurate mapping of forest cover and detection of tree cuttings. Detailed multi-temporal reference datasets were collected based on airborne laser scanning (ALS)-derived canopy height models (CHM) and very high-resolution (VHR) aerial orthomosaics. Reference datasets were used to train three machine learning (ML) models: random forest (RF), support vector machine (SVM), and feed-forward neural network (Nnet) in two forest districts located in Western and Northern Poland. The study also assessed the generalization capabilities of the best model in both local and temporal contexts. Regarding forest cover mapping, the OBIA RF classifier outperformed all other models with an overall accuracy (OA) of 99.27 % and Kappa of 98.18 %, while the PBIA SVM model showed the lowest (OA = 97.18 %, Kappa = 94.35 %). The testing of the model's generalization confirmed the performance of the OBIA RF model, with the Dice Coefficient ranging from 95.86 % to 96.74 %. The methodology's effectiveness in tree-cutting detection was demonstrated, with the detection rate ranging from 96.20 % to 99.39 % for the total number of cuttings, and from 99.45 % to 99.86 % for the total volume. In conclusion, the integration of PS satellite images, spectral-textural features, and generalized ML models proves to be effective for tree-cutting detection.

Deep and machine learning methods for the (semi-)automatic extraction of sandy shoreline and erosion risk assessment basing on remote sensing data (case of Jerba island -Tunisia)

Against the backdrop of the environmental crisis, the socio-economic, ecological and cultural importance of the coastal zone calls for greater awareness of how coastal resources function, evolve, are managed and enhanced. This study aims to develop a high-performance (semi-)automatic coastal monitoring method based on Landsat-5 and Sentinel-2 multispectral satellite images for spatiotemporal analysis of shoreline changes and erosion risk assessment along Jerba Island (Tunisia) using remote sensing data and geospatial tools. A comparative study between the band ratioing (BR) method and the pixel-based image analysis (PBIA) and object-based image analysis (OBIA) methods has led to the development of machine learning (ML), random forest (RF), deep learning (DL) and convolutional neural network (CNN) algorithms. Using these classification methods, 15 different shorelines were successively detected in 1989, 2015 and 2023 and then compared with a digitized reference shoreline from the Landsat-5 and Sentinel-2 images. Following a quantitative evaluation, the accuracy of the classification model shows that the combined CNN-OBIA approach provided the least accurate results, with an overall accuracy (OA) index of 67%, while the OBIA-RF classification method provided the most accurate results (OA of 95%). This comparative study identified an accurate and improved extraction method for quantifying changes in the position of the shoreline on the east coast of Jerba Island, enabling managers to make better decisions on coastal protection and adaptation to climate change.

Identification of Streamside Landslides with the Use of Unmanned Aerial Vehicles (UAVs) in Greece, Romania, and Turkey

The alleviation of landslide impacts is a priority since they have the potential to cause significant economic damage as well as the loss of human life. Mitigation can be achieved effectively by using warning systems and preventive measures. The development of improved methodologies for the analysis and understanding of landslides is at the forefront of this scientific field. Identifying effective monitoring techniques (accurate, fast, and low cost) is the pursued objective. Geographic Information Systems (GISs) and remote sensing techniques are utilized in order to achieve this goal. In this study, four methodological approaches (manual landslide delineation, a segmentation process, and two mapping models, specifically object-based image analysis and pixel-based image analysis (OBIA and PBIA)) were proposed and tested with the use of Unmanned Aerial Vehicles (UAVs) and data analysis methods to showcase the state and evolution of landslides. The digital surface model (DSM)-based classification approach was also used to support the aforementioned approaches. This study focused on streamside landslides at research sites in three different countries: Greece, Romania, and Turkey. The results highlight that the areas of the OBIA-based classifications were the most similar (98%) to our control (manual) classifications for all three sites. The landslides’ perimeters at the Lefkothea and Chirlesti sites showed similar results to the OBIA-based classification (93%), as opposed to the Sirtoba site, where the perimeters of the landslides from OBIA-based classification were not well corroborated by the perimeters in the manual classification. Deposition areas that extend beyond the trees were revealed by the DSM-based classification. The results are encouraging because the methodology can be used to monitor landslide evolution with accuracy and high performance in different regions. Specifically, terrains that are difficult to access can be surveyed by UAVs because of their ability to take aerial images. The obtained results provide a framework for the unitary analysis of landslides using modern techniques and tools.

Dynamic Mapping of Inland Freshwater Aquaculture Areas in Jianghan Plain, China

Freshwater aquaculture in Jianghan Plain plays a key role in the whole industry of China. It was expanding rapidly to meet the fast-growing demand of consumption in these decades. The spatial distribution change of aquaculture in Jianghan Plain has attracted many researchers in recent years and it is worth further investigating. However, the accuracy and the quality of inland aquaculture classification and mapping still have space to be improved. Our study attempts to use Sentinel-1 and Sentinel-2 data to classify aquaculture areas, non-aquaculture water, and non-water areas. We applied image segmentation and pixel-based feature computation to obtain candidate datasets. Recursive feature elimination (RFE) was used to find the optimal datasets which were then imported into a modified UNet for classification. This workflow was then applied to the annual datasets for better observation of the spatial change of aquaculture areas. In total 12 indices and features were found most influential and preserved after RFE. The overall classification accuracy reached 96.83%, while the precision and recall of aquaculture areas reached 84.47% and 90.48%, respectively. The combination of object-based and pixel-based image analysis showed its advantage in inland aquaculture classification compared to other studies. Integrated rice–crawfish farming was found a key factor motivating the rapid development of aquaculture. Policies about the fishing ban and ecosystem restoration were also found deeply affecting the spatial change of aquaculture in Jianghan Plain.

Geospatial-based machine learning techniques for land use and land cover mapping using a high-resolution unmanned aerial vehicle image

Bangladesh is primarily an agricultural country where technological advancement in the agricultural sector can ensure the acceleration of economic growth and ensure long-term food security. This research was conducted in the south-western coastal zone of Bangladesh, where rice is the main crop and other crops are also grown. Land use and land cover (LULC) classification using remote sensing techniques such as the use of satellite or unmanned aerial vehicle (UAV) images can forecast the crop yield and can also provide information on weeds, nutrient deficiencies, diseases, etc. to monitor and treat the crops. Depending on the reflectance received by sensors, remotely sensed images store a digital number (DN) for each pixel. Traditionally, these pixel values have been used to separate clusters and classify various objects. However, it frequently generates a lot of discontinuity in a particular land cover, resulting in small objects within a land cover that provide poor image classification output. It is called the salt-and-pepper effect. In order to classify land cover based on texture, shape, and neighbors, Pixel-Based Image Analysis (PBIA) and Object-Based Image Analysis (OBIA) methods use digital image classification algorithms like Maximum Likelihood (ML), K-Nearest Neighbors (KNN), k-means clustering algorithm, etc. to smooth this discontinuity. The authors evaluated the accuracy of both the PBIA and OBIA approaches by classifying the land cover of an agricultural field, taking into consideration the development of UAV technology and enhanced image resolution. For classifying multispectral UAV images, we used the KNN machine learning algorithm for object-based supervised image classification and Maximum Likelihood (ML) classification (parametric) for pixel-based supervised image classification. Whereas, for unsupervised classification using pixels, we used the K-means clustering technique. For image analysis, Near-infrared (NIR), Red (R), Green (G), and Blue (B) bands of a high-resolution ground sampling distance (GSD) 0.0125m UAV image was used in this research work. The study found that OBIA was 21% more accurate than PBIA, indicating 94.9% overall accuracy. In terms of Kappa statistics, OBIA was 27% more accurate than PBIA, indicating Kappa statistics accuracy of 93.4%. It indicates that OBIA provides better classification performance when compared to PBIA for the classification of high-resolution UAV images. This study found that by suggesting OBIA for more accurate identification of types of crops and land cover, which will help crop management, agricultural monitoring, and crop yield forecasting be more effective.

Deep learning for land use and land cover classification from the Ecuadorian Paramo.

ABSTRACT The paramo, plays an important role in our ecosystems as They balance the water resources and can retain substantial quantities of carbon. This research was carried out in the province of Tungurahua, specifically the Quero district. The aim is to develop a classification of the land use land cover (LULC) in the paramo using satellite imagery using several classifiers and determine which one obtains the best performance, for which three different approaches were applied: Pixel-Based Image Analysis (PBIA), Geographic Object-Based Image Analysis (GEOBIA), and a Deep Neural Network (DNN). Various parameters were used, such as the Normalized Difference Vegetation Index (NDVI), the Bare Soil Index (BSI), texture, altitude, and slope. Seven classes were used: paramo, pasture, crops, herbaceous vegetation, urban, shrubrainland, and forestry plantations. The data was obtained with the help of onsite technical experts, using geo-referencing and reference maps. Among the models used the highest-ranked was DNN with an overall precision of 87.43%, while for the paramo class specifically, GEOBIA reached a precision of 95%.

EVALUATING PIXEL-BASED AND OBJECT-BASED APPROACHES FOR FOREST ABOVE-GROUND BIOMASS ESTIMATION USING A COMBINATION OF OPTICAL, SAR, AND AN EXTREME GRADIENT BOOSTING MODEL

Abstract. The above-ground biomass (AGB) estimation monitoring provides a powerful tool for the assessment of carbon emission and sequestration. Using remote sensing technique is an environmentally friendly way of biomass estimation. Thus, this paper investigated optical (i.e. Landsat 8 OLI and Sentinel-2), synthetic aperture radar (SAR) (global phased array type L-band SAR (PALSAR/PALSAR-2) and Sentinel-1), and their integration for AGB estimation of the Pack demonstration forest in Warrensburg, NY. Importantly, a LiDAR AGB raster of the study area was used as reference data for training/testing purposes. Then, an extreme gradient boosting (Xgboost) machine learning model was used to predict biomass values. The major goal of this study was to compare the performance of pixel-based and object-based image analysis (OBIA) for the AGB estimation. Results indicated that the object-based approach improved the RMSE of AGB prediction about 6.28 Mg/ha for optical + SAR, 6.17 Mg/ha for SAR, and 5.6 Mg/ha for optical data in comparison to the pixel-based approach. Moreover, the combination of optical and SAR data increased the prediction accuracy regardless of feature extraction approach.

Decision tree-based machine learning models for above-ground biomass estimation using multi-source remote sensing data and object-based image analysis

Forest above-ground biomass (AGB) estimation provides valuable information about the carbon cycle. Thus, the overall goal of this paper is to present an approach to enhance the accuracy of the AGB estimation. The main objectives are to: 1) investigate the performance of remote sensing data sources, including airborne light detection and ranging (LiDAR), optical, SAR, and their combination to improve the AGB predictions, 2) examine the capability of tree-based machine learning models, and 3) compare the performance of pixel-based and object-based image analysis (OBIA). To investigate the performance of machine learning models, multiple tree-based algorithms were fitted to predictors derived from airborne LiDAR data, Landsat, Sentinel-2, Sentinel-1, and PALSAR-2/PALSAR SAR data collected within New York’s Adirondack Park. Combining remote sensing data from multiple sources improved the model accuracy (RMSE: 52.14 Mg ha−1 and R2: 0.49). There was no significant difference among gradient boosting machine (GBM), random forest (RF), and extreme gradient boosting (XGBoost) models. In addition, pixel-based and object-based models were compared using the airborne LiDAR-derived AGB raster as a training/testing sample. The OBIA provided the best results with the RMSE of 33.77 Mg ha−1 and R2 of 0.81 for the combination of optical and SAR data in the GBM model.

Identification and Counting of European Souslik Burrows from UAV Images by Pixel-Based Image Analysis and Random Forest Classification: A Simple, Semi-Automated, yet Accurate Method for Estimating Population Size

Burrowing mammals such as European sousliks are widespread and contribute significantly to soil ecosystem services. However, they have declined across their range and the non-invasive estimation of their actual population size has remained a challenge. Results support that the number of burrow entrances is positively correlated with population abundance, and burrow locations indicate the occupied area. We present an imagery-based method to identify and count animals’ burrows semi-automatically by combining remotely recorded red, green, and blue (RGB) images, pixel-based imagery, and random forest (RF) classification. Field images were collected for four colonies, then combined and processed by histogram matching and spectral band normalization to improve the spectral distinctions among the categories BURROW, SOIL, TREE, and GRASS. The accuracy indexes of classification for BURROW kappa (κ) were 95% (precision) and 90% (sensitivity). A 10-iteration bootstrapping of the final model resulted in coefficients of variation (CV%) of BURROW κ for sensitivity and precision lower than 5%; moreover, CV% values were not significantly different between those scores. The consistency of classification and balanced precision and sensitivity confirmed the applicability of this approach. Our approach provides an accurate, user-friendly, and relatively simple approach to count the number of burrow openings, estimate population abundance, and delineate the areas of occupancy non-invasively.

Seawall detection in Florida coastal area from high-resolution imagery using machine learning and OBIA

A methodology and framework were presented to detect the seawalls accurately and efficiently in low coastal areas and were then evaluated in the study area of Hallandale Beach City, Broward County, Florida. Aerial images collected by the Florida Department of Transportation were processed using eCognition Developer software for multiresolution segmentation and classification of objects. Two classification approaches, pixel-based image analysis and the object-based image analysis method, were applied for image classification. However, pixel-based classification was discarded for having less accuracy in output. Three techniques within object-based classification-machine learning (ML) technique, knowledge-based (KB) technique, and ML followed by KB technique were used to compare the most efficient method of classification. While performing the ML technique, three algorithms, such as random forest, support vector machine, and decision tree, were applied to test the best algorithm. Of all the approaches used, the combination of ML and a KB method was able to map the seawall effectively with an overall accuracy of 94%.

Spatio-temporal Mapping of Cotton Blooms Appearance Using Deep Learning

In this paper, we present an approach to create spatio-temporal maps using deep learning to visualize cotton bloom appearance over time. Specifically, we manually annotate cotton flower image data and train three state-of-the-art fast deep neural network models to count cotton blooms and their frequency over time prior to harvesting. We use the detection results of the best performing model combined with traditional pixel-based image analysis methods to create a map of where past and future blooms grow on a mid-stage cotton plant. The training results of our best model show a visual understanding of how many cotton flowers grow with high F1 Score of more than 0.95, a true positive rate of 98%, false negative and false positive rates both under 10%, and millisecond-scale inference time for real-time processing.

Machine learning and multi-sensor data fusion for mapping lithology: A case study of Kowli-kosh area, SW Iran

Lithological maps derived from traditional pixel-based image analysis are widely established in remote sensing. Recent development of new segmentation-based approaches such as object-based image analysis (OBIA) has led to significant improvements of accuracy in land cover classifications. However, additional enhancements have been reported by coupling OBIA with machine learning (ML) techniques. ML which automates computer algorithms and model building, enables learning, reasoning, and decision making through analysis and interpretation of data patterns and structures. This study is an attempt for evaluating and comparing the performances of four ML classifiers, including support vector machine (SVM), normal Bayes (NB), k-nearest neighbor (k-NN), and random forest (RF). This approach was tested with medium resolution imagery using an object-based classification procedure in a metamorphic complex, SW Iran. The multi-band input datasets were assembled form individual and multi-sensor layers from ASTER, Landsat 8 OLI, Sentinel-1 and Sentinel-2A. Results revealed that combined data sets obtain more reasonable classes for lithology, and combination of ASTER + simulated panchromatic of Sentinel-2A data shows the most efficient results. We also showed that, in general, RF and SVM algorithms were the most advantageous classifiers when using an object-based image analysis approach. When integrating the data sets aiming improvement of spatial resolution, the maximum accuracies of RF, SVM, k-NN and NB algorithms increased from 90% to 98%, 81% to 97%, 52 to 75%, and 40% to 64%, respectively. We also revealed that the integration of Sentinel-1 and Sentinel-2A data was promising and allowed for additional extraction of rock relief.

Performance of Sentinel-2A Remote Sensing System for Urban Area Mapping in Malaysia via Pixel-Based and OBIA Methods

Sentinel-2A remote sensing satellite system was recently launched, providing free global remote sensing data similar to Landsat systems. Although the mission enables the acquisition of 10 m spatial resolution global data, the assessment of Sentinel-2A data performance for mapping in Malaysia is still limited. This study aimed to investigate and assess the capability of Sentinel-2A imagery in mapping urban areas in Malaysia by comparing its performance against the established Landsat-8 data as well as the fusion datasets from combining Landsat-8 and Sentinel-2A datasets and using Wavelet transform (WT), Brovey transform (BT) and principal component analysis. Pixel-based and object-based image analysis (OBIA) classification approaches combined with support vector machine (SVM) and decision tree (DT) algorithms were utilized in this assessment, and the accuracy generated was analysed. The Sentinel-2A data provided superior urban mapping output over the use of Landsat-8 alone, and the fusion datasets do not yield advantages for single-scene urban mapping. The highest overall accuracy (OA) for pixel-based classification of Sentinel-2A images is 84.77 % by SVM, followed by 65.27 % using DT. BT produced the highest OA for the fusion images of 78.40 % with SVM and 52.21 % with DT. For the object-based classification of Sentinel-2A images, the highest OA is 71.33 % by SVM, followed by 76.38 % using DT. Similarly, the highest OA of fusion images is obtained by BT of 50.35 % with SVM, followed by 65.66 % with DT. From the analysis, the use of SVM pixel-based classification for medium spatial resolution Sentinel-2A data is effective for urban mapping in Malaysia and useful for future long-term mapping applications. HIGHLIGHTS An accurate mapping of urban land is still challenging and requires high image quality of spectral and spatial aspects to identify features Single and fusion image analysis conducted in order to investigate and assess the most performing interpretation result by grouping out the features classes Statistical performance and image classification comparison is relevant to prove the most effective result among the images GRAPHICAL ABSTRACT

Building Change Detection Method to Support Register of Identified Changes on Buildings

Based on a newly adopted “Rulebook on the records of identified changes on buildings in Serbia” (2020) that regulates the content, establishment, maintenance and use of records on identified changes on buildings, it is expected that the geodetic-cadastral information system will be extended with these records. The records contain data on determined changes of buildings in relation to the reference epoch of aerial or satellite imagery, namely data on buildings: (1) that are not registered in the real estate cadastre; (2) which are registered in the real estate cadastre, and have been changed in terms of the dimensions in relation to the data registered in the real estate cadastre; (3) which are registered in the real estate cadastre, but are removed on the ground. For this purpose, the LADM-based cadastral data model for Serbia is extended to include records on identified changes on buildings. In the year 2020, Republic Geodetic Authority commenced a new satellite acquisition for the purpose of restoration of official buildings registry, as part of a World Bank project for improving land administration in Serbia. Using this satellite imagery and existing cadastral data, we propose a method based on comparison of object-based and pixel-based image analysis approaches to automatically detect newly built, changed or demolished buildings and import these data into extended cadastral records. Our results, using only VHR images containing only RGB and NIR bands, showed object identification accuracy ranging from 84% to 88%, with kappa statistic from 89% to 96%. The accuracy of obtained results is satisfactory for the purpose of developing a register of changes on buildings to keep cadastral records up to date and to support activities related to legalization of illegal buildings, etc.

VEGETATION EXTRACTION WITH PIXEL BASED CLASSIFICATION APPROACH IN URBAN PARK AREA

Information on urban vegetation and land use is critical for sustainable environmental management in cities. In general, urban vegetation is important for urban planning because it helps to maintain a balance between the natural environment and the built-up region. The assessment of the composition and configuration of the vegetation is important to highlight the urban ecosystem. Thus, obtaining information about urban vegetation is critical for developing a sustainable urban development strategy. Remote sensing is increasingly being used to generate such data for mapping and monitoring changes in urban vegetation. The aim of this study is to identify and classify vegetation using the high-resolution Pleiades satellite image in urban park areas using pixel-based image analysis. Pixel based method was applied and support vector machine algorithm was used for classification of urban vegetation. Comparison of accuracy was made from the error matrices, overall accuracy and kappa coefficient for vegetation and non-vegetation classes. The overall accuracy for the classification approach was 98.98% and a kappa value of 0.97. The result demonstrates the ability of high-resolution imagery to accurately extract urban vegetation despite the complex surface of the urban area. The findings can be used to support other research and applications related to urban green space monitoring, conservation, and future urban vegetation planning.

Comparison of PBIA and GEOBIA classification methods in classifying turbidity in reservoirs

Our goal is to compare the performance of Classification and Regression Tree, Naive Bayes and Random Forest algorithms, from supervised image classification, and approaches on Pixel-Based Image analysis (PBIA) and Geographic Object-Based Image Analysis (GEOBIA), to classify turbidity in reservoirs. Tod do so, we use Landsat 8 image and bands and spectral indices, as predictive parameters, as well as the classification algorithms based on PBIA and GEOBIA. The Brazilian Itaipu reservoir was adopted, as a case study. Our results show that the RF classifier obtained the highest accuracy in both classification approaches, followed by CART and NB. The KA and OA indices of the GEOBIA classifications were superior to the PBIA classifications in both algorithms. This study contributes with an approach to quickly and accurately delineating turbidity spectral limits in reservoirs.

Microphysiological System for High-Throughput Computer Vision Measurement of Microtissue Contraction.

The ability to measure microtissue contraction in vitro can provide important information when modeling cardiac, cardiovascular, respiratory, digestive, dermal, and skeletal tissues. However, measuring tissue contraction in vitro often requires the use of high number of cells per tissue construct along with time-consuming microscopy and image analysis. Here, we present an inexpensive, versatile, high-throughput platform to measure microtissue contraction in a 96-well plate configuration using one-step batch imaging. More specifically, optical fiber microprobes are embedded in microtissues, and contraction is measured as a function of the deflection of optical signals emitted from the end of the fibers. Signals can be measured from all the filled wells on the plate simultaneously using a digital camera. An algorithm uses pixel-based image analysis and computer vision techniques for the accurate multiwell quantification of positional changes in the optical microprobes caused by the contraction of the microtissues. Microtissue constructs containing 20,000-100,000 human ventricular cardiac fibroblasts (NHCF-V) in 6 mg/mL collagen type I showed contractile displacements ranging from 20-200 μm. This highly sensitive and versatile platform can be used for the high-throughput screening of microtissues in disease modeling, drug screening for therapeutics, physiology research, and safety pharmacology.

Evaluating pixel-based vs. object-based image analysis approaches for lithological discrimination using VNIR data of WorldView-3

The object-based against pixel-based image analysis approaches were assessed for lithological mapping in a geologically complex terrain using Visible Near Infrared (VNIR) bands of WorldView-3 (WV-3) satellite imagery. The study area is Hormuz Island, southern Iran, a salt dome composed of dominant sedimentary and igneous rocks. When performing the object-based image analysis (OBIA) approach, the textural and spectral characteristics of lithological features were analyzed by the use of support vector machine (SVM) algorithm. However, in the pixel-based image analysis (PBIA), the spectra of lithological end-members, extracted from imagery, were used through the spectral angle mapper (SAM) method. Several test samples were used in a confusion matrix to assess the accuracy of classification methods quantitatively. Results showed that OBIA was capable of lithological mapping with an overall accuracy of 86.54% which was 19.33% greater than the accuracy of PBIA. OBIA also reduced the salt-and-pepper artifact pixels and produced a more realistic map with sharper lithological borders. This research showed limitations of pixel-based method due to relying merely on the spectral characteristics of rock types when applied to high-spatial-resolution VNIR bands of WorldView-3 imagery. It is concluded that the application of an object-based image analysis approach obtains a more accurate lithological classification when compared to a pixel-based image analysis algorithm.

Impacts of light detection and ranging (LiDAR) data organization and unit of analysis on land cover classification

ABSTRACT Airborne light detection and ranging (LiDAR) data have been used to generate land cover models for almost two decades. In this paper, three common processing decisions are assessed for their impact on the accuracy and configuration of the resultant land cover models. Using data acquired from a single-wavelength, discrete return system, this study compares six land cover models that investigate (i) the organization of data into tiles or flightstrips, (ii) the unit of analysis as either the individual LiDAR point or as a pixel in a rasterized model of the LiDAR data, and (iii) the use of either pixel- or object-based image analysis. Although the overall accuracies of the land cover models generated in this study are comparable, models disagree on up to 17% of the total study area. Class-specific metrics of recall and precision differ markedly between models, and the configuration of land covers are also affected. Models that employ pixel-based image analysis techniques tend to generate models with smaller, more dispersed patches of land cover. Data organization and choice of unit of analysis also influence the configuration of land cover, although effects differ depending on the land cover class. Comprehensive analyses of accuracy and precision are crucial to developing land cover models. This study demonstrates that it is also important to understand the potential influence of classification methodologies on the configuration of landscape features, especially when interpreting land cover models from an ecological or landscape genetic perspective.